axosomatic synapses on granule cells are preserved in ... 04-04 sere… · granule cells of the...
TRANSCRIPT
357)
© Charles University in Prague – The Karolinum Press, Prague 2004
Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Axosomatic Synapses on Granule Cellsare Preserved in Human Non-infiltratingTumour or Lesion-related and MesialTemporal Sclerotic Epilepsy, but MarkedlyReduced in Tumour-infiltrated DentateGyrus with or without EpilepsySeress L.
1, Ábrahám H.
1, Dóczi T.
2, Pokorny J.
3, Klemm J.
4, Bakay R. A.
4
1Central Electron Microscopic Laboratory of Faculty of Medicine, University
of Pécs, Hungary;
2Department of Neurosurgery of Faculty of Medicine, University of Pécs, Hungary;
3Institute of Physiology of the First Faculty of Medicine, Charles University in
Prague, Czeh Republic;
4Department of Neurosurgery, Rush Presbyterian-St. Luke’s Medical Center,
Chicago, Illinois, USA
Received December 2, 2004, Accepted December 29, 2004
Abstract: Granule cells of the human hippocampal dentate gyrus were examined.
In controls, granule cells displayed somatic spines and cell nuclei with small
infoldings. In addition, the cytoplasm of human granule cells always displayed
lipofuscin. Subsurface cisterns of endoplasmic reticulum were frequently observed
in the human granule cells. Two types of axosomatic synapses were found; most
frequently symmetric and less frequently asymmetric. Many of the axosomatic
synapses were isolated by glial processes in tumour or lesion-related epileptic
patients, but the ultrastructural characteristics of granule cells were not different
from those of the control patients. Large bundles of reactive astroglial fibres
appeared regularly in all layers of the dentate gyrus. In tumour infiltrated
hippocampi, glial processes dominated the neuropil and the number of perisomatic
synapses was markedly reduced. Reduction in the number of perisomatic synapses
did not correlate with severity and duration of seizures but did correlate with the
malignancy of the tumour. It is suggested that reduction of perisomatic inhibition
may not be a characteristic of granule cells in the epileptic human dentate gyrus.
Key words: Hippocampus – Perisomatic inhibition – Epilepsy – Neuropathology.
Supported by the Hungarian Scientific Research Fund (OTKA) grant #T047109.
Mailing address: Prof. László Seress, MD., PhD., Central Electron Microscopic
Laboratory, Faculty of Medicine, University of Pécs, Szigeti u. 12., 7643 Pécs,
Hungary, Phone: +36 72 536 060, e-mail: [email protected]
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
358) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Introduction
It is generally accepted that the axon terminals form exclusively symmetric
synapses with the somata of glutamatergic pyramidal cells of the cerebral cortex
[1, 2], whereas on the cell bodies of GABAergic local circuit neurones both
symmetric and asymmetric synapses occur [3, 4]. Although, this general principle is
valid for the archicortical hippocampal formation, there is one exception. Granule
cells of the dentate gyrus are glutamatergic but display both symmetric and
asymmetric axosomatic synapses [5]. Granule cells of the dentate gyrus form a
homogeneous group in rodents, but they display significant variability in the
primate hippocampus [6]. When compared to rodents, granule cells of non-human
primates display a low number of axosomatic synapses [7]. Parvalbumin-positive
basket cells, which provide the main inhibitory synaptic input for the somata
of granule cells, are low in number in the monkey and human dentate gyrus
[8, 9, 10]. Similarly, the number of perisomatic symmetric synapses is similarly low
in the human as it is in the monkey [11].
Perisomatic inhibition is regarded to be crucial for defending the cell from
overexcitation and a decrease of perisomatic GABAergic inhibition was considered
as a cause of focal epilepsy in rodent animal models [12]. A decreased number of
perisomatic inhibitory synapses may be considered to be pathognomic for human
epilepsy, although a recent study demonstrated an unchanged number of
perisomatic synapse for the dentate granule cells in mesial temporal sclerotic
epilepsy [11].
In the present study, the frequency and distribution of axosomatic synapses as
well as the ultrastructural characteristics of granule cells are described in the
human hippocampal dentate gyrus of controls and epileptic patients.
Materials and Methods
Human “control” hippocampal tissue was obtained from three cases (Table 1).
In cases 1 and 3, coroner’s autopsy was performed and brains were removed two
hours after death. The brain was perfused through the carotid arteries first with
physiological saline followed by a fixative containing 4% paraformaldehyde. In
control case 2, the hippocampus was placed into the 4% paraformaldehyde
fixative immediately after the surgical removal. From all control hippocampi,
10-mm-thick slices were cut and postfixed with 2.5% glutaraldehyde for 2h at
room temperature. Then blocks were treated as described below for the surgically
removed tissue.
Axosomatic synaptic density on granule cells of mesial temporal sclerotic
epilepsy patients (Tab. 1) was examined on electron microscopic preparations
obtained and previously described by Franck and co-workers (12). Patients had
intractable mesial temporal epilepsy with different degrees of hippocampal atrophy
from mild gliosis (case 1161) to hippocampal sclerosis (cases 1225, 1227). In the
previous study, only the biocytin-filled granule cells were examined and synapses
Perisomatic Inhibition in Human Epileptic Dentate Gyrus
359)Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
were identified only on the labelled granule cells. In the present study, axosomatic
synaptic coverage of randomly selected well-preserved non-labelled granule cells
was examined inside the granule cell layer.
Axosomatic synaptic density on granule cells of tumour-related and tumour-
infiltrated epileptic patients (Tab. 1) was counted in preparations obtained from
surgically removed hippocampi. The growing tumours of the temporal neocortex
either compressed the hippocampal region and caused herniation of the uncus or
infiltrated the hippocampus. Mild or moderate gliosis was observed, but
hippocampal sclerosis was not found in these cases.
The surgically removed brain tissue was immersed in 4% paraformaldehyde
buffered with phosphate buffer (PB, 0.1 M, pH 7.4). After transportation to the
histological laboratory, the hippocampi were cut in 10-mm-thick blocks immersed
in 4% paraformaldehyde and kept for 2-4 h at room temperature. Then the blocks
were cut with Vibratome at 60 or 80 µm. Sections were placed in 2.5%
glutaraldehyde solution for 2h at room temperature, washed several times with
PB, osmicated with 1% OsO4 for 1 h, dehydrated and flat-embedded in Durcupan
according to routine electron microscopic procedures. Flat embedded sections
were examined under the light microscope. Areas of interest were selected and
re-blocked in Durcupan. Thin sections were collected on Parlodion-coated one-
slot grids, and stained with lead citrate and uranyl acetate. Sections were examined
with a JEOL 1200 electron microscope. A continuous row of 80–120 granule cells
could be examined on each one-slot grid in each of the thin section. A goniometer
was used to tilt the grids for the identification of synapses as well as to determine
the type of synapses. When a synapse (synaptic density, vesicle accumulation,
Table 1
Group Case Age (years) Gender Diagnose Seizure onset
Control Control 1. 47 cardiac arrest no seizure
Control 2. 48 lung cancer with no seizure
intracranial metastasis
Control 3. 65 cardiac arrest no seizure
Mesial temporal JF 1161 43 CPS 23 years
epipeltic cases JF 1227 43 SPS, CPS 28 years
JF 1225 44 CPS 34 years
Tumour and Case No. 11. 45 meningeoma 1.5 years
lesion-related Case No. 12. 46 glioblastoma multiforme 6 months
epileptic cases Case No. 17. 19 angioma 3 years
Case No. 18. 61 glioblastoma multiforme 6 months
Tumour- Case No. 1 65 glioblastoma no seizure
infiltrated Case No. 28 17 glioma 2 years
hippocampus Case No. 15 11 glioblastoma no seizure
TU/00/4 50 glioma 4 months
Abbreviations: SPS, simple partial seizure; CPS, complex partial seizure
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
360) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
synaptic cleft) could not equivocally be identified than the apposing terminal was
not counted in this study.
Results
The granule cell layer of the control human dentate gyrus was densely packed
with granule cells (Fig. 1A and B). Somata of granule cells directly apposed each
other without having intervening glial processes (Fig. 2A). The thin cytoplasm
frequently contained lipofuscin pigment deposits, but never formed large
aggregates (Fig. 2A and B). Somal spines (Fig. 2C) and small nuclear indentations
were commonly found in granule cells. Somal spines varied in both diameter and
length, but lacked polyribosomes or granular endoplasmic reticulum. Subsurface
cisterns were frequently observed under the plasma membrane and they
preferentially occurred in the vicinity of synaptic terminals (Fig. 2D). Somata of
granule cells were apposed by axon terminals that formed symmetric (Fig. 1C)
or asymmetric synapses (Fig. 1D). The average number of symmetric axosomatic
synapses per granule cell per thin section was 0.44 (Tab. 2). The number of
synapses found on a single granule cell soma per thin section ranged from 0 to 3.
About 65% of granule cells displayed no synapses on their somal surface in the
examined thin section. The percentage of asymmetric axosomatic synapses was
6–10% (Tab. 2) and most of them were on granule cells located at the border of
the molecular layer.
Diameter of somata of granule cells could be measured from semithin section in
the light microscope (Fig. 1B) and also from thin sections in the electron
microscope using a calibration bar drawn on the screen of the electron
microscope. A population of at least 100 neurones was measured in the electron
microscope for each case except for the mesial temporal sclerotic epileptic cases,
where the quality of sections did not allow the examination of such a large
population, because sections were prepared from tissue slices. The mean of
shorter and longer diameters was calculated in those cases where the soma was
elliptical. From the diameter, the circumference of the soma was calculated. For
example, in control case No.1, the average length of somatic membrane for a
granule cell was 39.72 µm. The length was calculated from the average of somal
diameters that was 12.65 ± 0.135 µm (S.E.M.) as measured in the electron
microscope. The average nuclear diameter was 9.5 ± 0.138 µm (S.E.M.) and the
nucleoli were 2 µm long in diameter. In average, the synaptic density was
1 synapse for 85–100 µm long somal membrane.
In mesial temporal sclerotic epileptic cases there was no significant difference in
the number or size of the axosomatic synapses (Tab. 3) when compared to the
controls. The somal features of granule cells including the lipofuscin deposits and
submembrane cisterns were also similar as found in the controls. Astroglial
processes were more frequent in mesial temporal epileptic cases than in controls,
but glial processes did not separate granule cells.
Perisomatic Inhibition in Human Epileptic Dentate Gyrus
361)Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Figure 1 – A. Photomicrograph of cresyl violet stained 10-µm-thick section of the granule cell (gl) layer and
B. Photomicrograph of toluidine blue stained semi-thin section of the granule cell layer (gl) from the
hippocampus of control case 2. Double-headed arrows indicate the somatic diameter as it was measured
to calculate the length of somatic perimeter. C and D. Electron micrographs of axon terminals (t) that form
symmetric (C) or asymmetric (D) synapses (arrows) with somata of granule cells (g) in the dentate gyrus
of control case No.1. Bar = 10 µm for A and B; 0.2 µm for C and D.
A B
C D
glgl
g
g
g
t
t
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
362) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Figure 2 – Electron micrographs of different features of granule cells (Control case No.1). A. Apposing
somata of granule cells (g) without having glial sheets among them. The cytoplasm (g) contains lipofuscin
(l). B. Multiple large lipofuscin deposits (l) regularly appear in granule cells. C. Granule cells (g) regularly
display somal spines (s) and well developed subsurface cisterns (arrow) that appear to be continuous with
the endoplasmic reticulum (D). These cisterns frequently locate in the vicinity of synapses or where an axon
terminal (t) apposes the soma. Bar = 0.5 µm for A and B and 0.2 µm for C and D.
A B
C D
g
g
I
I
I
g
g
S
t
g
Perisomatic Inhibition in Human Epileptic Dentate Gyrus
363)Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
In the non-infiltrating lesion or tumour-related cases (Tab. 1) the neocortical
tumours did not infiltrate the hippocampus as was verified by the pathological
examination. The growing tumours caused life-threatening compression of the
neighbour brain tissue including the hippocampal formation. In all cases, seizures
of variable duration were recorded (Tab. 1). Independent of the length of epileptic
history before the operation, perisomatic innervation of granule cells of the
surgically removed hippocampal dentate gyrus did not differ from that of the
control. Frequency, shape and size of axon terminals were similar as found in the
controls (Tab. 4, Fig. 4A). However, the granule cell layer was loosely packed with
granule cells (Fig. 3A) and glial processes were frequent among granule cells
(Fig. 4A). Astroglial processes surrounded axon terminals, but did not disrupt the
synaptic contact (Fig. 4A). The abundant filament content of astroglial processes
suggests that all of these processes belonged to activated glial cells. The majority of
glial processes run perpendicular to the granule cell layer and most of the somata
of activated glial cells were in the hilus. Separation of somata of granule cells varied
considerably inside the same granule cell layers, because in some places only small
profiles (1µm) of cross-sections of glial processes appeared, whereas a few
hundred microns away, somata of other granule cells were completely separated
by glial fiber bundles (Fig. 4A).
In the tumour-infiltrated cases, all layers of dentate gyrus were occupied by
tumour cells (Fig. 3B). As a result, granule cells were separated not only by glial
processes, but also by the somata of tumour cells. Neuropil of the dentate gyrus
was disrupted by the glial processes forming thick bundles among granule cells
(Fig. 4B). Correspondingly, the number of axosomatic synapses was low in all of
these cases, especially in the heavily infiltrated cases (Fig. 3B, 4B) where the
number of synapses was one-tenth of the control value (Tab. 5). Interestingly, the
low number of synapses did not correlate with seizure frequency, because similarly
low numbers were found in those cases where no seizures were recorded (Tab. 1).
Discussion
The results of this study suggest that human granule cells display characteristics
similar to those of the non-human primate in respect of their somal spines, nuclear
infoldings and the number of axosomatic synapses [7]. In addition, a commonly
Table 2 – Number of axosomatic synapses per granule cell in controls
Number of synapses Percentage
Number (Number of terminals Synapses / cell of asymmetric
of cells forming synapses) Mean ± S.E.M. synapses (%)
Control 1. 243 112 (108) 0.46 ± 0.044 18
Control 2. 235 87 (84) 0.38 ± 0.053 16
Control 3. 190 90 (89) 0.47 ± 0.075 10
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
364) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Figure 3 – Light microscopic photomicrographs of the granule cell layer in epileptic patients. A. Granule
cells are not as closely packed in the epileptic granule cell layer (gl) of the dentate gyrus as in controls
(compare with A and B on Fig. 1). The molecular layer (ml) and hilus (h) has the regular density of cells
similarly as seen in controls. A tumour-related case of epilepsy, where a neocortical meningeoma
compressed the hippocampal region. B. The granule cell layer (gl) as well as the molecular layer (ml) and
hilus (h) are densely packed by tumour cells. The space among granule cells is large therefore the cellular
density in granule cell layer (gl) appears to be much smaller than in controls. Dentate gyrus of an epileptic
patient where the ganglioglioma infiltrated the hippocampus. Bar = 25 µm for A and B.
A
B
ml
gl
h
ml
gl
h
Perisomatic Inhibition in Human Epileptic Dentate Gyrus
365)Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
Figure 4 – Electron micrographs of the granule cell layer from epileptic patients. A. Granule cells (g) are
separated by glial processes (f). The glial fibres appear to isolate axon terminals (t) but do not destroy them
in the tumour-related epilepsy. Same case as shown on Fig. 3A. B. Wide bands of glial processes (f)
separate granule cells (g) in the dentate gyrus of epileptic patient where the tumour infiltrated the dentate
gyrus. Same case as shown on Fig. 3B. Bar = 0,5 µm for A and B.
A
B
g
g
f
f
g
f
g
f
f
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
366) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
observed feature of human granule cells was the occurrence of lipofuscin pigment
granules. Lipofuscin has been described for the human pyramidal and non-
pyramidal hippocampal neurones, but not for granule cells [14, 15]. It appears that
in human lipofuscin occurs in all hippocampal neurones and the size, location and
amount of pigment granules are characteristic for the neuronal type.
Another interesting feature was the frequent occurrence of subsurface cisterns.
Subsurface cisterns have been known to occur in neurone [16]. Indeed, such
Table 3 – Number of axosomatic synapses per granule cell in mesial
temporal epileptic cases
Percentage
Number Number Synapses / cell of asymmetric
of cells of synapses Mean ± S.E.M. synapses (%)
JF 1161. 50 13 0.26 ± 0.032 –
JF 1225. 50 18 0.36 ± 0.050 5
JF 1227. 50 24 0.48 ± 0.055 8
Table 4 – Number of axosomatic synapses per granule cell in tumour-
related epilepsy (Cases 11, 12, 18) and lesion-related epilepsy (Case 17)
Percentage
Number Number Synapses / cell of asymmetric
of cells of synapses Mean ± S.E.M. synapses (%)
Case No. 11. 100 36 0.36 ± 0.055 not counted
Case No. 12. 100 39 0.39 ± 0.062 not counted
Case No. 17. 100 46 0.46 ± 0.041 not counted
Case No. 18. 154 87 0.57 ± 0.071 7
Table 5 – Number of axosomatic synapses per granule cell in epileptic
cases with tumour-infiltrated hippocampus
Percentage
Number Number Synapses / cell of asymmetric
of cells of synapses Mean ± S.E.M. synapses (%)
Lightly infiltrated
Case No. 1. 150 19 0.18 ± 0.036 not counted
Case No. 28. 190 17 0.21 ± 0.032 17
Heavily infiltrated
Case No. TU/00/4. 187 19 0.05 ± 0.004 11
Case No. 15. 107 19 0.08 ± 0.005 11
Perisomatic Inhibition in Human Epileptic Dentate Gyrus
367)Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
cisterns exist in pyramidal neurones of Ammon’s horn, but less frequently than in
granule cells. Subsurface cisterns are not characteristic for the human granule cells
only, because they were also found in rodent granule cells (unpublished
observation). Such cisterns are known to participate in the Ca++
homeostasis of
the cell [17]. It is not known whether these cisterns contain calcium-binding
proteins, although they have a strategic location close to synapses to accumulate
free Ca++
.
The granule cells of the human dentate gyrus displayed a number of axosomatic
synapses similar to those found in rhesus monkeys [7]. The majority of the
axosomatic synapses (90%) was of symmetric type, similarly to those in monkeys
[7]. A previous study demonstrated a similarly low number (1.2/cell) of axosomatic
synapses on human granule cells [18]. Identification of synapses is difficult in non-
perfused tissues; therefore the difference between the absolute values can be
explained with the methods used. In the present study, synapses were identified
directly in the microscope, whereas Wittner et al. [18] used a computer assisted
high-resolution digital camera system. Neither Wittner et al. [18] nor our study
could demonstrate a significant change in the average number of symmetric
synapses in mesial temporal sclerotic epileptic patients. A previous finding
suggested that most of axosomatic synapses on the somata of granule cells in
human mesial temporal sclerotic epileptic hippocampi were of asymmetric type
[12]. However, in that study only a few biocytin-labelled granule cells were
examined in the electron microscope. In biocytin-labelled cells, the DAB reaction
frequently interferes with a clear identification of the type of synapses. In this
study, we have not found significant change in the percentage of asymmetric
axosomatic synapses neither in mesial temporal sclerotic epilepsy nor in non-
infiltrating tumour-related epilepsy. Our recent findings support the results of
previous studies, showing no decrease in the number of GABAergic neurones and
symmetric axosomatic terminals in the epileptic dentate gyrus [11, 18].
In the tumour-infiltrated dentate gyrus, the number of axosomatic synapses was
markedly reduced even in those cases when the tumour infiltration was moderate,
suggesting that processes of neoplastic glial cells separate axon terminals from
their postsynaptic targets, whereas the reactive astroglia fibres only isolate the
synaptic axon terminals. Interestingly, the reduction in axosomatic innervation did
not correlate with onset of seizures, because similar loss of terminals occurred in
epileptic and non-epileptic tumour-infiltrated cases. These results strengthen the
view that a change in perisomatic innervation of granule cells has no direct
relationship with the onset of epilepsy.
Acknowledgements
The authors wish to thank dr. Philip A. Schwartzkroin for kindly providing the
electron microscopic material prepared by late JoAnn E. Franck whose helpful and
kind personality will always be remembered by her former colleagues.
Seress L., Ábrahám H., Dóczi T., Pokorny J., Klemm J., Bakay R. A.
368) Prague Medical Report / Vol. 105 (2004) No. 4, p. 357–368
References
1. COLONNIER M.: Synaptic patterns of different cell types in the different laminae of the cat visual
cortex. Brain Res. 9: p. 268–287, 1968.
2. PARNAVELA S J. G., SULLIVAN K., LIEBERMAN A. R., WEBSTER K. E.: Neurones and their synaptic
organization in the visual cortex of the rat. Electron microscopy of Golgi preparations. Cell. Tiss. Res.
83: p. 499–517, 1977.
3. SOMOGYI P., KISVARDAY Z. F., MARTIN K. A. C., WHITTERIDGE D.: Synaptic connections of
morphologically identified and physiologically characterized large basket cells in the striate cortex of
cat. Neuroscience 10: p. 261–295, 1983.
4. HENDRY S. H. C., HOUSER C. R., JONES E. G., VAUGHN, J. E.: Synaptic organization of
immuncytochemically identified GABA neurones in the monkey sensory-motor cortex. J. Neurocytol.
12: p. 639–660, 1983.
5. SERESS L., RIBAK C. E.: A substantial number of asymmetric axosomatic synapses is a characteristic
of the granule cells of the hippocampal dentate gyrus. Neurosci. Letters. 56: p. 21–26, 1985.
6. SERESS L., FROTSCHER M.: Morphological variability is a characteristic feature of granule cells in the
primate fascia dentata: a combined Golgi/electron microscopic study. J. Comp. Neurol. 293: p. 253–267,
1990.
7. SERESS L., RIBAK C. E.: Ultrastructural features of primate granule cell bodies show important
differences from those of rats: axosomatic synapses, somatic spines and infolded nuclei. Brain Res. 569:
p. 353–357, 1992.
8. RIBAK C. E., NITSCH R., SERESS, L.: Proportion of parvalbumin-positive basket cells in the
GABAergic innervation of pyramidal and granule cells of the rat hippocampal formation. J. Comp.
Neurol. 300: p. 449–461, 1990.
9. SERESS L., GULYAS A. I., FREUND, T. F.: Parvalbumin- and calbindin D28k-immunoreactive neurones
in the hippocampal formation of the macaque monkey. J. Comp. Neurol. 313: p.162–177, 1991.
10. SERESS L., GULYAS A. I., FERRER I., TUNON T., SORIANO E., FREUND, T. F.: Distribution,
morphological features, and synaptic connections of parvalbumin- and calbindin D28k-immunoreactive
neurones in the human hippocampal formation. J. Comp. Neurol. 337: p. 208–230, 1993.
11. MAGLOCZKY, ZS., WITTNER, L., BORHEGYI, ZS., HALA SZ, P., VAJDA, J. CZIRJAK, S.,
FREUND, T. F.: Changes in the distribution and connectivity of interneurons in the epileptic human
dentate gyrus. Neuroscience 96: p. 7–25, 2000.
12. FRANCK J. E., POKORNY J., KUNKEL D. D., SCHWARTZKROIN P. A.: Physiologic and morphologic
characteristics of granule cell circuitry in human epileptic hippocampus. Epilepsia 36: p. 543–558, 1995.
13. RIBAK C. E.: Epilepsy and the Cortex. In: Cerebral Cortex. Vol. 9. Peters A., Jones E.G. (eds.) Plenum
Press, 1991, p. 427–483.
14. MAGLOCZKY ZS., HALASZ P., VAJDA J., CZIRJAK S., FREUND T. F.: Loss of calbindin-D28k
immunoreactivity from dentate granule cells in human temporal lobe epilepsy. Neuroscience 76:
p. 377–385, 1997.
15. OLBRICH H. G., BRAAK H.: Ratio of pyramidal cells versus non-pyramidal cells in sector CA1 of the
human Ammon’s horn. Anat. Embryol. 173: p. 105–110, 1985.
16. ROSENBLUTH J.: Subsurface cisterns and their relationship to the neuronal plasma membrane. J. Cell
Biology 13: p. 405–421, 1962.
17. MCBURNEY R. N., NEERING I. R.: Neuronal calcium homeostasis. TINS 10: p. 164–169, 1987.
18. WITTNER L., MAGLOCZKY ZS., BORHEGYI ZS., HALASZ P., TOTH SZ., EROSS L, SZABO Z.,
FREUND T. F.: Preservation of perisomatic inhibitory input of granule cells in the epileptic human
dentate gyrus. Neuroscience 108: p. 587–600, 2001.